Development and implementation of a real time machine learning based approach for detecting phishing websites
- Author
- Muzanargo, Dempssey
- Title
- Development and implementation of a real time machine learning based approach for detecting phishing websites
- Abstract
- This undertaking offers a plugin for detecting phishing for the Chrome browser that detects and notifies users concerning phishing websites in real time using a random forest classifier. According to the author's review of the literature, the random forest classifier outperforms other algorithms in detecting phishing websites. One frequent method consists of executing the categorization on a server and then enable the plugin to query the server for the results. In contrast to this strategy, our project proposes to carry out the categorization within the browser. The advantages of classifying in the client side browser include increased privacy (the user's data from browsing does not need to leave his computer) and phishing detection is not affected by network latency. The proposed system will accurately predict phishing websites in the browser and prevent users from sharing sensitive information with potential hackers. The proposed system will use JavaScript so that it can run as a browser plugin. Because JavaScript does not support many ML libraries, and taking into account the processing capability of the client machines, the strategy must be lightweight. The model must be trained using Python scikit-learn on the phishing websites dataset, and the model parameters must then be exported into a portable format for usage in JavaScript.
- Date
- June 2023
- Publisher
- BUSE
- Keywords
- Phishing Websites
- Supervisor
- Mr. O. Muzurura
- Item sets
- Department of Computer Science
- Media
-
Muzanargo, Dempssey.pdf